Student Behavior Recognition From Heterogeneous View Perception in Class Based on 3-D Multiscale Residual Dense Network for the Analysis of Case Teaching.

Hui Liu,Xia Wu,Yang Liu,Ran Zhang

doi:10.3389/fnbot.2021.675827

Abstract

The study of student behavior analysis in class plays a key role in teaching and educational reforms that can help the university to find an effective way to improve students' learning efficiency and innovation ability. It is also one of the effective ways to cultivate innovative talents. The traditional behavior recognition methods have many disadvantages, such as poor robustness and low efficiency. From a heterogeneous view perception point of view, it introduces the students' behavior recognition. Therefore, we propose a 3-D multiscale residual dense network from heterogeneous view perception for analysis of student behavior recognition in class. First, the proposed method adopts 3-D multiscale residual dense blocks as the basic module of the network, and the module extracts the hierarchical features of students' behavior through the densely connected convolutional layer. Second, the local dense feature of student behavior is to learn adaptively. Third, the residual connection module is used to improve the training efficiency. Finally, experimental results show that the proposed algorithm has good robustness and transfer learning ability compared with the state-of-the-art behavior recognition algorithms, and it can effectively handle multiple video behavior recognition tasks. The design of an intelligent human behavior recognition algorithm has great practical significance to analyze the learning and teaching of students in the class.

Highlights

A country is prosperous and strong when the education is strong
This paper focuses on the behavior recognition architecture based on a 3-D convolutional neural network (CNN). 3-D convolution can be used to extract universal and reliable space-time features directly from the original video, which is intuitive and effective
To solve the problem that the traditional 3-D CNN algorithm lacks full utilization of the network’s multilevel convolutional features, this paper proposes a 3-D multiscale residual dense network architecture for human behavior recognition and verifies the effectiveness of the proposed algorithm on public and real scene data sets

Summary

INTRODUCTION

A country is prosperous and strong when the education is strong. China has always attached great importance to the development of education. The proposed 3-D multiscale residual dense network (3DMRDNet) imitates the residual of ResNet learning and the dense DenseNet network connection mode to build the 3-D residual dense blocks and extract the multilevel space–time features in 3D video behavior recognition In the deep network structure, to ensure the maximum information flow between different levels in the network, the skip connection mode of the residual network is adopted in 3D-RDB, which connects feature graphs with the same feature map size so that the output of each layer is directly connected to the input of the subsequent layer This kind of jumping connection alleviates the problem of network gradient disappearance, enhances feature propagation, promotes feature reuse, and retains the features of forward propagation. Compared with 3D-ResNet and 3D-DenseNet models, 3D-MRDNet has the advantages of fewer parameters and less computations

EXPERIMENTS AND ANALYSIS

Method

CONCLUSIONS

Findings

ETHICS STATEMENT